ChemicalScience

EDGE ARTICLE

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.

View Article OnlineView Journal | View Issue

Isomer-depende

aLaboratory for Synchrotron Radiation and

5232 Villigen, Switzerland. E-mail: patrick.hbInstitute for Chemical and Bioengineering

Biosciences, ETH Zurich, 8093 Zurich, SwitzcLaboratory for Catalysis and Sustainable

Villigen, Switzerland

† Electronic supplementary informa10.1039/d1sc00654a

Cite this: Chem. Sci., 2021, 12, 3161

All publication charges for this articlehave been paid for by the Royal Societyof Chemistry

Received 2nd February 2021Accepted 5th February 2021

DOI: 10.1039/d1sc00654a

rsc.li/chemical-science

© 2021 The Author(s). Published by

nt catalytic pyrolysis mechanismof the lignin model compounds catechol,resorcinol and hydroquinone†

Zeyou Pan, ab Allen Puente-Urbina, b Andras Bodi, a Jeroen A. vanBokhoven bc and Patrick Hemberger *a

The catalytic pyrolysis mechanism of the initial lignin depolymerization products will help us develop

biomass valorization strategies. How does isomerism influence reactivity, product formation, selectivities,

and side reactions? By using imaging photoelectron photoion coincidence (iPEPICO) spectroscopy with

synchrotron radiation, we reveal initial, short-lived reactive intermediates driving benzenediol catalytic

pyrolysis over H-ZSM-5 catalyst. The detailed reaction mechanism unveils new pathways leading to the

most important products and intermediates. Thanks to the two vicinal hydroxyl groups, catechol (o-

benzenediol) is readily dehydrated to form fulvenone, a reactive ketene intermediate, and exhibits the

highest reactivity. Fulvenone is hydrogenated on the catalyst surface to phenol or is decarbonylated to

produce cyclopentadiene. Hydroquinone (p-benzenediol) mostly dehydrogenates to produce p-

benzoquinone. Resorcinol, m-benzenediol, is the most stable isomer, because dehydration and

dehydrogenation both involve biradicals owing to the meta position of the hydroxyl groups and are

unfavorable. The three isomers may also interconvert in a minor reaction channel, which yields small

amounts of cyclopentadiene and phenol via dehydroxylation and decarbonylation. We propose

a generalized reaction mechanism for benzenediols in lignin catalytic pyrolysis and provide detailed

mechanistic insights on how isomerism influences conversion and product formation. The mechanism

accounts for processes ranging from decomposition reactions to molecular growth by initial polycyclic

aromatic hydrocarbon (PAH) formation steps to yield, e.g., naphthalene. The latter involves a Diels–Alder

dimerization of cyclopentadiene, isomerization, and dehydrogenation.

Introduction

Lignin, one of the main biomass components, can be convertedto fuels and ne chemicals by pyrolysis.1–3 Due to its variedfunctionalization and complex structure, even state-of-the-artapproaches suffer from limited selectivity and controlla-bility.4–7 To optimize conversion and improve selectivity ina targeted way, it is necessary to understand the chemistry ofearly lignin pyrolysis products. Studies on representative ligninbuilding blocks, such as guaiacol, syringol and eugenol, canprovide detailed insight into the (catalytic) pyrolysis mecha-nism.8–12 Li et al. used electron paramagnetic resonance (EPR)spectroscopy to show that guaiacol pyrolysis is a radical-drivenprocess and catechol is a favored product due to the presence of

Femtochemistry, Paul Scherrer Institute,

emberger@psi.ch

, Department of Chemistry and Applied

erland

Chemistry, Paul Scherrer Institute, 5232

tion (ESI) available. See DOI:

the Royal Society of Chemistry

mobile hydrogen atoms.8 Verma and Kishore carried out densityfunctional theory (DFT) calculations and showed that thelowest-energy unimolecular decomposition pathway of eugenolyields guaiacol.9 Detailed mechanistic studies exist for benzal-dehyde,13 phenol,14 dimethoxybenzenes15 and larger modelsystems, such as 4-phenoxyphenol and 2-methoxy-phenoxybenzene.16

Benzenediols, and especially catechol (1,2-benzenediol, 1o),represent basic structural units in the lignin network, are majorintermediates or products in lignin thermal decomposition,and stand out because of their reactivity, thanks to the presenceof two hydroxyl groups. A detailed mechanistic description ofbenzenediol pyrolysis would be of great value to describe relatedderivatives and construct models to predict their behavior inlignin decomposition. Thus, it is of fundamental interest to

Chem. Sci., 2021, 12, 3161–3169 | 3161

Chemical Science Edge Article

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.View Article Online

elucidate the pyrolysis mechanism fully by focusing on the roleof isomerism and the change in the mechanism in the presenceof a zeolite catalyst.

Yang et al. investigated catechol, resorcinol, and hydroqui-none (o-, m- and p-benzenediol) pyrolysis in a two-stage tubularreactor, analyzed the products by online gas chromatographyand reported isomer-dependent product distributions. The p-benzoquinone product was only observed for hydroquinone (p-benzenediol), while resorcinol (m-benzenediol) producedsignicantly more CO2 and C5 hydrocarbons than the other twoisomers.17 Based on computations, a biradical reaction pathwaywas proposed.18 By using in situ photoelectron photoion coin-cidence (PEPICO) methods, we found compelling spectroscopicevidence (vide infra) that the resorcinol decomposition mecha-nism is dominated by a retro-Diels–Alder reaction to produceCO2 and C5H6 isomers (R1) instead or it leads to two highlyreactive ketenes species (R1).18,19 Furthermore, the three ben-zenediols share the same reaction to form cyclopenta-2,4-dien-1-ol (C5H6O) and 2,4-cyclopentadiene-1-one (C5H4O), initiatedby CO loss in a phenolic keto–enol tautomerism pathway (seeR2 for an overview).

The pyrolysis mechanism of catechol was investigated byOrmond et al. using matrix isolation infrared spectroscopy andiPEPICO detection.20 Fulvenone, a highly reactive intermediate,was produced only in catechol, because of the vicinal hydroxylgroups opening up the dehydration reaction R3.

They proposed that the phenolic keto–enol tautomerismpathway of catechol, i.e., carboxyl group formation, followed byCO and H losses, may also apply to resorcinol and hydroqui-none (see R2),20 as conrmed later for resorcinol.19 Based onthese considerations, we can clearly expect (i) different reactionpathways for the three benzenediol isomers which can only berevealed by (ii) in situ and operando detection methods.

How does the reactivity, the intermediates, and the productschange in the presence of a catalyst? How does the catalystinuence the benzenediol pyrolysis depending on the arenesubstitution pattern? Yang et al. pyrolyzed catechol with andwithout H-ZSM-5 at 550–950 �C and reported that the catalystlowers the reaction temperature and increases the amount ofaromatic hydrocarbons produced via deoxygenation. However,they did not discuss mechanistic insights in detail.21 We haveinvestigated the catalytic pyrolysis of guaiacol and found cate-chol as a demethylation product. It dehydrates to the highlyreactive fulvenone (see R3), which either forms cyclopentadienevia decarbonylation and hydrogenation, or phenol via radical

3162 | Chem. Sci., 2021, 12, 3161–3169

pathways. Thus, the fulvenone ketene was identied as thecentral intermediate.22 Ketenes have recently been in the spot-light because of their crucial role in numerous catalyticprocesses.23 Although most studies observe catechol in lignincatalytic pyrolysis, isomer-dependent reaction mechanismsremain elusive, probably owing to the limited time resolutionand/or isomer specicity of conventional analysis methods.24–27

Imaging photoelectron photoion coincidence (iPEPICO) spec-troscopy is a multiplexed, sensitive and isomer-selective tech-nique28 that allows us to identify reactive intermediates andderive a reaction mechanism in complex and reactive systemsranging from combustion and pyrolysis to catalysis.19,20,22,29–31

This is achieved by utilizing photoion mass-selected thresholdphotoelectron spectra (ms-TPES) for isomer-selective detectionof the key intermediates. Reactive intermediates and productsare soly ionized using vacuum ultraviolet (VUV) synchrotronradiation, producing photoions and photoelectrons that aredetected in delayed coincidence. In addition to mass spectra,the photoelectron spectrum associated to individual m/z peakscan be plotted for chemical analysis. This represents a powerfultool to detect and assign reactive and short-lived intermediatesin benzenediol catalytic pyrolysis isomer-specically, asdetailed in the ESI.†31

This study aims to understand the chemistry of H-ZSM-5-catalyzed pyrolysis of the benzenediols catechol, resorcinoland hydroquinone. Specically, we want to elucidate howisomerism inuences the pyrolytic conversion, reactive inter-mediates, and the reaction mechanism to yield aromatics withand without a zeolite catalyst. These insights will lay the foun-dation to study even more complex lignin model compoundsand to control selectivity and conversion in biomassvalorization.

Results and discussionGas chromatography–mass spectrometry

Fig. 1 shows the benzenediol product assignment aer catalyticpyrolysis in a batch-type reactor by gas chromatography–massspectrometry (py-GC/MS) based on the NIST08 mass spectrumlibrary. The detailed assignment is shown in the ESI† in TablesS1–S3.† Based on the integral peak intensities in the ion chro-matograms (Fig. S1†), catechol has the highest reactivity andlargest conversion. Besides phenol, cyclopentadiene andbenzene, polycyclic aromatic hydrocarbons (PAHs) such asnaphthalene, indene and their alkyl-substituted derivativesdominate the spectra. These are the products of secondaryreactions, including decomposition, methylation, isomeriza-tion processes and molecular growth, due to the high concen-tration and long residence time of the reactants, intermediates,and products in the reactor. Reactive species, such as fulvenone(m/z 92), 2-cyclopenten-1-one (m/z 82) and ketene (m/z 42), evadedetection because of their short lifetime and the GC/MS detec-tion limits. While resorcinol shows similar products as cate-chol, it does so at a much lower conversion (see Fig. S1†).Hydroquinone has a higher selectivity towards phenol and p-benzoquinone with fewer and less intense contributions oflarger aromatics.

© 2021 The Author(s). Published by the Royal Society of Chemistry

Fig. 1 py-GC/MS results of catechol (red sticks, top), resorcinol (bluesticks, center) and hydroquinone (green sticks, bottom) obtained at530 �C. Besides small molecular and oxygen-containing products,monocyclic and polycyclic aromatic hydrocarbons, i.e., coke precur-sors, are observed.

Edge Article Chemical Science

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.View Article Online

Overall, py-GC/MS gives a detailed picture of the nalpyrolysis products, such as phenol and PAHs, but does not allowfor insights into the initial stages that govern the whole reactionmechanism. Thus, we investigated the pyrolysis chemistry withand without catalyst using our in situ PEPICO setup.

Fig. 2 Time-of-flight mass spectra recorded at hn ¼ 10.5 eV uponnon-catalytic (light-colored) vs. catalytic (dark-colored) fast pyrolysisof catechol (red trace, top), resorcinol (blue trace, center) andhydroquinone (green trace, bottom). Catalyst: H-ZSM-5. *Acetoneimpurity in the chamber background.

Photoelectron photoion coincidence spectroscopy

Catalytic benzenediol decomposition. A continuous injec-tion molecular beam experiment, including a sample containerand quartz reactor (see Scheme S1†), was used to extend thelifetime of the reactive intermediates and investigate the initialreaction steps. The solid sample is vaporized in the containerand mixes with the carrier gas. It reacts on the catalyst surfacein the quartz reactor, packed with glass wool, at ca. 100 mbartotal pressure. Products and intermediates exiting the catalyticpyrolysis reactor expand into high vacuum, form a molecularbeam, are skimmed and then ionized by VUV synchrotronradiation from the Swiss Light Source (Paul Scherrer Institute).32

The ions and electrons are accelerated in a constant electriceld and detected by velocity map imaging detectors to measurephotoionization mass spectra. Peaks in the mass spectra wereisomer-selectively identied by their photoion mass-selected

© 2021 The Author(s). Published by the Royal Society of Chemistry

threshold photoelectron spectra (ms-TPES) combined withadiabatic ionization energy (AIE) calculations33 and Franck–Condon (FC) simulations,34 or reference spectra.

First, catalytic and non-catalytic pyrolysis can be comparedwith the help of the spectra shown in Fig. 2 and S2.† The massspectra of catechol, resorcinol, and hydroquinone are relativelyclean in non-catalytic pyrolysis (Fig. 2, light-colored traces), andonly few products are observed at temperatures as high as580 �C. While we observe m/z 80 in catechol and resorcinol,which can be assigned to cyclopentadienone (c-C5H4]O, seeR2), we see exclusively p-benzoquinone (m/z 108) in hydroqui-none. These ndings agree with unimolecular decompositionproducts reported earlier.19,20

Upon adding H-ZSM-5 catalyst, the conversion increasessignicantly, especially for catechol (see Fig. 2, dark-coloredtraces), in agreement with the observations of Yang et al.,21 whilethe other two isomers show notably increased reactivity only athigher temperatures. A more detailed temperature comparison oncatalytic and non-catalytic pyrolysis for catechol, resorcinol andhydroquinone is presented in Fig. S2.† The differences in conver-sion among the isomers become even more pronounced at lowtemperatures and if the three isomers are directly compared at thesame conditions, as shown in Fig. S3.†

The intermediates and products in the mass spectra wereanalyzed via ms-TPE spectroscopy and the most important

Chem. Sci., 2021, 12, 3161–3169 | 3163

Fig. 3 Representative ms-TPE spectra of products along with FCsimulations or reference spectra.35

Chemical Science Edge Article

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.View Article Online

assignments are depicted in Fig. 3. A direct comparison of thems-TPES between the benzenediols is given in Fig. S4 and TableS4 in the ESI.† Catechol (m/z 110, Fig. 2 dark red trace) hasa relatively high selectivity towards m/z 66, which is assigned tocyclopentadiene (see Fig. 3). Furthermore, we assigned fulveneand benzene (both m/z 78), 2,4-cyclopentadiene-1-one (m/z 80),2-cyclopenten-1-one (m/z 82), phenol (m/z 94), fulvenone (m/z92) and naphthalene (m/z 128), as intermediates and productsof the catechol catalytic pyrolysis.

The second most reactive benzenediol is hydroquinone (m/z110, Fig. 2 dark green trace). The main contributors to the massspectrum are the pyrolysis products cyclopentadiene (m/z 66),phenol (m/z 94) and p-benzoquinone (m/z 108), which areobserved at higher temperatures compared to catechol. Theproduct distribution of resorcinol (m/z 110, Fig. 2 dark bluetrace) resembles that of catechol but exhibits the lowestconversion among the three benzenediols in the completetemperature range studied. While cyclopentadiene was onlyobserved in the gas-phase pyrolysis of resorcinol,19,20 it is seen inall three benzenediols over H-ZSM-5, which is clearly a dis-tinguishing feature of catalytic pyrolysis. This raises the possi-bility that there is benzenediol isomerization on the catalystsurface. Although most of the assigned intermediates andproducts are shared among the three benzenediols, we onlyfound fulvenone in high concentrations in the catechol exper-iments (see m/z 92 ms-TPES in Fig. S4†). In the 8.5–10.0 eVphoton energy range, the fulvenone TPES is dominated bytransitions into the strongly coupled ground and rst elec-tronically excited state of the cation.35 The m/z 92 ms-TPESbarely rises above the background in the resorcinol andhydroquinone experiments, which precludes assignment(Fig. S4†). In such cases, we can rely on the photoionization (PI)spectrum, which includes all ionization events leading to them/z channel of interest, not only the threshold ones. The m/z 92 PIspectra in resorcinol and hydroquinone catalytic pyrolysis are

3164 | Chem. Sci., 2021, 12, 3161–3169

virtually identical (Fig. S5†) and agree perfectly with the fulve-none reference spectrum below 8.7 eV.

Starting from 8.9 eV, the spectrum rises more steeply thanthe fulvenone reference, which indicates a signicant toluenecontribution. This proves that the three benzenediols catalyti-cally interconvert, because only the ortho isomer, catechol, candehydrate to yield fulvenone, thanks to the vicinal hydroxylgroups (see R3).20 In the absence of a catalyst, these isomeri-zations are observed neither in previous pyrolysisexperiments19,20 nor herein (see Fig. 2). Benzenediolinterconversion is further veried by the detection of p-benzoquinone (m/z 108), the dehydrogenation product ofhydroquinone, also in the resorcinol and catecholexperiments (Fig. S4†). In contrast to fulvenone, p-benzoquinone is a stable compound that could have beendetected by py-GC/MS of resorcinol and catechol, but it wasnot, probably because of the detection limit of this technique.Therefore, benzenediol catalytic isomerization is a catalytic sidechannel compared with the decomposition processes.

The catechol mechanism can be described as follows(Scheme 1): fulvenone was shown to be the key species in theformation of cyclopentadiene (m/z 66) and phenol (m/z 94) fromcatechol in the catalytic fast pyrolysis of guaiacol.22 In brief,catechol (1o) is dehydrated in an acid-catalyzed reaction on thecatalyst surface, yielding fulvenone (30, Scheme 1), which issubsequently hydrogenated to form C6H5O radicals (31 and 32),which decarbonylate and hydrogenate to produce cyclo-pentadiene (6) or hydrogenate to phenol (22).22 A second ben-zenediol deoxygenation reaction channel is initiated bydecarbonylation to access ve-membered ring species, such ashydroxyl cyclopentadienes (3) and 2-cyclopenten-1-one (4, seeR2). This reaction is also observed in non-catalytic pyrolysis butat comparably higher temperatures and was explored byquantum chemical calculations in resorcinol.19,20 In addition,benzenediols (1) may dehydroxylate to yield phenol (22) andwater in a Brønsted acid catalyzed reaction. Furthermore,hydroquinone (1p) can dehydrogenate to form p-benzoquinone(20), as summarized in Scheme 1. It is worth noting that ben-zenediol dehydrogenation reaction is always endothermic, asseen by the CBS-QB3-calculated reaction enthalpies for R4–R6:

While o- and p-benzoquinone (C6H4O2) formation are simi-larly endothermic, the formation of the biradical meta-isomerrequires twice as much energy and is, thus, less likely to occur

© 2021 The Author(s). Published by the Royal Society of Chemistry

Scheme 1 H-ZSM-5-catalyzed pyrolysis reaction mechanism of the three benzenediol isomers. Thick arrows represent the dominant reactionpathways, leading to the most abundant species.

Edge Article Chemical Science

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.View Article Online

during the catalytic pyrolysis of resorcinol. o-Benzoquinonemayindeed be thermodynamically accessible, but not detected,which can be explained by its high reactivity towardsdecarbonylation.15

These insights into the early reaction pathways are valuable.However, we found them to be lacking to fully understand themechanism leading to the catalytic pyrolysis products observedin GC21 or in py-GC/MS experiments herein. Especially, thedehydroxylation and PAH formation mechanisms seem elusive,which are nevertheless crucial in lignin conversion. We there-fore decided to dig deeper and reveal the chemistry of theprompt pyrolysis products more in detail.

Intermediate product chemistry. While condensation reac-tions are crucial in PAH-formation (see below), decompositionprocesses yield progressively smaller products and only abun-dantly available species on the catalysis surface may participate,such as hydrogen or, at most, methyl radicals.22 Thus, one canshed more light on the catalytic pyrolysis mechanism of ben-zenediols by investigating the sequential catalytic pyrolysisreactions of the prompt pyrolysis products individually andselectively. The mass spectra for the catalytic pyrolysis of p-benzoquinone (m/z 108), 2-cyclopenten-1-one (m/z 82), phenol(m/z 94), and cyclopentadiene (m/z 66) are shown in Fig. 4. Asbefore, the intermediates and products were assigned based ontheir ionization energy and threshold photoelectron spectrum(Fig. 3, S6–S9 and Table S4†). In the p-benzoquinone experiment(Fig. 4, purple trace), we identied phenol (m/z 94), 2-cyclopenten-1-one (m/z 82), 2,4-cyclopentadiene-1-one (m/z 80),

© 2021 The Author(s). Published by the Royal Society of Chemistry

fulvene (m/z 78), benzene (m/z 78), cyclopentadiene (m/z 66), 1,3-butadiene (m/z 54), ketene (m/z 42) and propene (m/z 42) aspyrolysis products. The product distribution agrees quite wellwith that of hydroquinone (Fig. 2 dark green trace). Thissuggests that hydroquinone (1p) is easily formed by hydroge-nation of p-benzoquinone (20) (see Scheme 1), due to theabundance of hydrogen on the H-ZSM-5 catalyst. As p-benzo-quinone is an important contributor to the hydroquinonespectra, as well, we can establish that hydroquinone and p-benzoquinone both readily desorb from the catalyst surface andthat there is an equilibrium between them thanks to fastinterconversion. This is not in contradiction with the stronglyendothermic dehydrogenation enthalpy for R6, because thehydrogen atoms are more strongly bound on the surface than ina hydrogen molecule, which decreases the effective dehydro-genation energy signicantly. In addition to the productsformed via hydroquinone, the direct decarbonylation of p-ben-zoquinone (C6H4O2) yields 2,4-cyclopentadiene-1-one (18,C5H4O), which can decarbonylate (see Scheme 1) again toproduce 1-buten-3-yne (19, C4H4) as shown in R7:

These reactions were also experimentally observed andcomputationally conrmed by Robichaud et al. during thepyrolysis of p-dimethoxybenzene, which intermediately yields p-benzoquinone.15 1-Buten-3-yne (19) is prone to hydrogenation

Chem. Sci., 2021, 12, 3161–3169 | 3165

Fig. 4 Time-of-flight mass spectra at hn ¼ 10.5 eV obtained uponcatalytic pyrolysis of p-benzoquinone (purple trace), 2-cyclopenten-1-one (green trace), phenol (blue trace) and cyclopentadiene (redtrace) over H-ZSM-5. *Acetone impurity in chamber background.

Chemical Science Edge Article

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.View Article Online

to yield 1,3-butadiene (15) on the catalyst surface (Scheme 1).Besides, it may also yield acetylene (12) and ethylene at elevatedtemperatures. PAHs (e.g. naphthalene and indene) are alsoobserved at elevated temperatures, analogous to our py-GC/MSresults. Their formation mechanism is discussed below.

The green trace in Fig. 4 shows the mass spectrum of 2-cyclopenten-1-one catalytic pyrolysis. We assign the massspectral peaks to 2,4-cyclopentadiene-1-one (m/z 80), 2-bute-none (m/z 70), cyclopentadiene (m/z 66), 1,3-butadiene (m/z 54),ketene (m/z 42), and propene (m/z 42) with the help of ms-TPEspectra (Fig. S7†). 2-Cyclopenten-1-one (4) can dehydrogenateon the catalyst surface to yield 2,4-cyclopentadiene-1-one (18) ordemethylate via ring opening to form 2-butenone (16) (seeScheme 1) aer hydrogenation over H-ZSM-5. The latter candecompose further to yield ketene (17) along with acetylene(12). The loose methyl groups on the catalyst surface can readilymethylate the products (6/ 7), similar to what was observed inguaiacol catalytic pyrolysis.22 Decarbonylation of 4 results in 1,3-butadiene (15). Furthermore, cyclopentadiene (6) was identiedin small quantities, which could tentatively be produced bya sequence of Brønsted acid catalyzed hydrogenation anddehydroxylation reactions of 2-cyclopenten-1-one (4 / 6).

3166 | Chem. Sci., 2021, 12, 3161–3169

Probing deoxygenation channels. Phenol catalytic pyrolysisshows a much simpler picture and reveals the dehydroxylationreaction mechanism of phenolics. Benzene (m/z 78) and naph-thalene (m/z 128) are the major products, while ethylene (m/z28), propene (m/z 42), and ketene (m/z 42) are observed only insmall quantities (Fig. 4 blue trace). The dehydroxylation ofphenol on the catalyst surface (Scheme 1, 22 / 23 / 9) yieldsbenzene together with only trace amounts of fulvene (m/z 78ms-TPES in Fig. S8†). This is a crucial nding as it points to twoformation channels of C6H6 (vide infra). In addition, phenol (22)can isomerize to 2,4-cyclohexadienone (24) via phenolic keto-enol tautomerization, which decarbonylates to cyclo-pentadiene (6, Scheme 1) as observed also during non-catalyticpyrolysis and conrmed via quantum chemical calculations byScheer et al. (R8):36

Cyclopentadiene (6) is only observed in very low concentra-tions (see Fig. S10†) during phenol catalytic pyrolysis due to itshigh reactivity (vide infra).

While these measurements have lled in the gaps in themechanism, they have not revealed details of aromatic growthand PAH formation, a crucial step in coking and catalyst deac-tivation. Indeed, the ms-TPES of m/z 128 shows contributionsfrom 1-buten-3-ynylbenzene and/or 3-buten-1-ynylbenzene (seeFig. S9†) beside naphthalene, as veried by FC simulations andpoints towards complex PAH chemistry. For instance, phenylradicals (23) could recombine with acetylene to naphthalene(29) in hydrogen abstraction–acetylene addition (HACA)37–40

(Scheme 1, 23/ 28/ 29), but neither the 28 intermediate norethynylbenzene were observed. 23, a non-resonantly stabilizedradical, may require higher activation energy, and the pathway(23 / 28 / 29) may only play a minor role. In addition, wefound traces of m/z 152 (Fig. S10†), which can tentatively beassigned as ethynyl naphthalene or acenaphthylene.40,41 Thus,a more detailed investigation of the early PAHs and cokeintermediates is necessary to unlock their formationmechanism.

Probing aromatic growth

The catalytic pyrolysis of cyclopentadiene, presented in Fig. 4(red trace), offers valuable insights into PAH and early cokeformation.42 Interestingly, cyclopentadiene completely dis-appeared from the spectrum at the highest temperature, 582 �C(Fig. S11† le). This also explains why it is only present in traceamounts in phenol catalytic pyrolysis (Fig. 4 blue trace). Thepropene (m/z 42) peak in cyclopentadiene catalytic pyrolysisconrms that cyclopentadiene is a source of C3 species by ringopening to produce propene and acetylene/ethylene.14 C3

moieties, such as allyl (C3H5) and propargyl (C3H3) radicals, arebenzene precursors par excellence, as was recently found inpropane oxyhalogenation over CrPO4.43 However, the low pro-pene signal points towards other PAH formation channels inthe phenol experiments.43 It is well known that cyclopentadiene

© 2021 The Author(s). Published by the Royal Society of Chemistry

Scheme 2 Formation of polycyclic aromatic compounds via dimer-ization of cyclopentadiene and subsequent dehydrogenation or C–Cbond cleavage.

Fig. 5 ms-TPE spectra ofm/z 132 in dicyclopentadiene at 21 �C and itscatalytic pyrolysis at 389 �C. Isomerization reactions to tetrahy-dronaphthalenes47 are clearly observed.

Edge Article Chemical Science

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.View Article Online

(6) can dimerize and dehydrogenate to naphthalene (29), asshown in Schemes 1 and 2.44,45 Besides decomposition anddimerization, cyclopentadiene (6) may also grow by formaladdition of CH3 units, yielding fulvene (8), benzene (9) andtoluene (10). The peak at m/z 80 could be assigned to a combi-nation of three methyl-cyclopentadiene isomers (Fig. S9†),which implies methylation on the catalyst and likely contributesto fulvene formation via dehydrogenation (6 / 7 / 8).22

Interestingly, large contributions (Fig. 4 red trace) of naphtha-lene (29), 1-buten-3-ynylbenzene, 3-buten-1-ynylbenzene to m/z 128and of tetrahydronaphthalenes (C10H12, 26, 27) and dicyclopenta-diene (C10H12, 25) tom/z 132 are found in ourms-TPES analysis (seeFig. S9†). Assuming that their ionization cross sections are similar,it is eye-catching that the bicyclic C10H12 isomers have a compa-rable if not higher abundance in the mass spectrum than themonocyclic toluene and benzene/fulvene (Fig. 4 red trace). As theC5H6 dimer was not observed at room temperature and in theabsence of the catalyst, it must be formed during the reaction ofcyclopentadiene on H-ZSM-5. Thus, it is conceivable that thespecies above m/z 66 are produced via dimerization of cyclo-pentadiene, isomerization, and subsequent decomposition tofulvene/benzene and C4 (25/ 9) species or toluene and C3 species(25/ 10), respectively.45 The latter chemistry is corroborated by theobservation of 1-buten-3-ynylbenzene or 3-buten-1-ynylbenzenebesides naphthalene (all m/z 128).

To verify this hypothesis, we investigated the catalyticpyrolysis of dicyclopentadiene, as shown in Fig. S11† right.The appearance of cyclopentadiene proves that the dimerundergoes a retro-Diels–Alder reaction on the catalyst surface,which is an endothermic process.46 The product distributionof dicyclopentadiene is similar to that of cyclopentadiene (Fig.S11† le), and naphthalene (m/z 128) appeared at elevatedtemperatures, too. In order to understand the dimerizationand formation of naphthalene, we measured the ms-TPES ofm/z 132 during H-ZSM-5-catalyzed pyrolysis of dicyclopenta-diene (Fig. 5). We found three isomers of the compositionC10H12 at 389 �C reactor temperature. Two isomers areassigned to 1,2,3,4- and 1,4,5,8-tetrahydronaphthalene (27,26), while the band at 8.7 eV is attributed to dicyclopentadiene(25) in both the exo and endo forms. This shows that dicyclo-pentadiene indeed isomerizes rst and then continually

© 2021 The Author(s). Published by the Royal Society of Chemistry

dehydrogenates via formal 2H2 loss reactions to yield naph-thalene or ring-opens to produce the two buten-ynylbenzenes(34 and 35). The possible reaction pathways of dicyclopenta-diene (25) in catalytic pyrolysis is presented in more detail inScheme 2 and shows strong evidence that cyclopentadiene isthe precursor for the formation of early coke precursors suchas naphthalene and indene. Our ndings are supported byexperiments of Sato et al., who reported that dicyclopenta-diene yields aromatics such as benzene, toluene, xylene,indene, naphthalene and coke.45 This further evidences thatC5 intermediates are mainly responsible for the PAH forma-tion in the catalytic pyrolysis of benzenediols and, possibly,larger lignin moieties, as well. These reactions likelycontribute to the hydrocarbon pool and thus to coke forma-tion, leading to catalyst deactivation.

Overview of benzenediol catalytic pyrolysis mechanism

We can now derive a reaction mechanism for the catalytic fastpyrolysis of benzenediols, summarized in Scheme 1. Catechol(1o) shows the highest conversion, due to the favorable dehy-dration of the adjacent hydroxyl groups to form fulvenone (30),a short-lived intermediate. Fulvenone is hydrogenated to C6H5Oradicals (31 and 32), which are either hydrogenated to yieldphenol (22) on the catalyst surface or decarbonylated andhydrogenated to form cyclopentadiene (6). Especially the latteris observed in large quantities at temperatures as high as500 �C. Fulvenone (30) is also a minor intermediate in thecatalytic pyrolysis of resorcinol (1m) and hydroquinone (1p),due to the minor catalytic isomerization channel of the twoisomers to catechol (1o), observed only in catalytic benzenedioldecomposition.14,20 However, benzenediol interconversion isa relatively slow process on H-ZSM-5 and does not contributesignicantly to the conversion of the less reactive isomers.48

All three isomers can be dehydroxylated in the presence ofa hydrogen source to yield phenol (22), which is one of the

Chem. Sci., 2021, 12, 3161–3169 | 3167

Chemical Science Edge Article

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.View Article Online

major products and only observed in catalytic pyrolysis. Owingto the large hydrogen abundance on H-ZSM-5, phenol (22) canbe further dehydroxylated catalytically to benzene (9) or itdecarbonylates to yield cyclopentadiene (6). Naphthalene (29)can be produced from cyclopentadiene via Diels–Alder cyclo-addition followed by isomerization to tetrahydronaphthalenes(26 and 27) and subsequent dehydrogenation. The third ben-zenediol decarbonylation pathway, yielding 2-cyclopenten-1-one (4), is the main reason for the wide distribution of prod-ucts. Not only does 2-cyclopenten-1-one (4) decompose toolenic species, such as 1-buten-3-yne (19) and 1,3-butadiene(15), but it also yields ketene (17) and produces cyclopentadiene(6) by a second dehydroxylation reaction. Subsequently, the C5

species generate C6 and C7 species either by dimerization ofcyclopentadiene (6) and subsequent loss of C4 or C3 species (25/ 10 and 25 / 9) or by consecutive methylation steps startingfrom C5 species. The observed methyl cyclopentadienes (7)dehydrogenate to fulvene (8), which is a benzene (9) precursor.Due to the presence of C2 and C3 species, other recombinationreactions are also conceivable, in which cyclopentadiene isultimately responsible for molecular growth to produce indene(25 / 33) or naphthalene (25 / 26/27 / 29) and coke at laterstages of the reaction, as depicted in Scheme 2.

Conclusions

Catechol, resorcinol, and hydroquinone have isomer-dependent reaction products in non-catalytic gas-phase pyrol-ysis. While 2-cyclopenten-1-one is a common product for thethree benzenediols, ethenone or fulvenone ketenes are onlyobserved in meta- and ortho-benzenediol, respectively. In cata-lytic pyrolysis, the reaction temperature could be signicantlydecreased, and the products are similar. Catechol has thehighest reactivity because of the vicinal hydroxyl groups, whichpromote dehydration and raise the selectivity towards fulve-none production. The detection of these highly reactive speciescan only be achieved utilizing sensitive, multiplexed, anduniversal detection tools like PEPICO with VUV synchrotronradiation. Fulvenone represents a shortcut from catechol tocyclopentadiene and phenol formation and explains the higherconversion for catechol. Resorcinol and hydroquinone, on theother hand, exhibit lower conversion but can still producecyclopentadiene via the 2-cyclopenten-1-one intermediate,which can tautomerize to hydroxyl cyclopentadiene. In addi-tion, we found a well-dened reaction pathway to polycyclicaromatic hydrocarbons, such as naphthalene and indene,formed via Diels–Alder addition of cyclopentadiene andsubsequent dehydrogenation. Our results show that there areseveral reaction pathways to cyclopentadiene, which is a PAHand coke precursor and is thus responsible for the deactivationof the catalyst. The catalytic pyrolysis mechanism is dominatedby deoxygenation and aromatic growth and intermediateproduct chemistry.

Benzenediols, especially catechol, are common structuralmoieties in lignin networks. Understanding their reactionmechanism provides the necessary insight to optimize the cata-lytic pyrolysis process in general. We have shown that

3168 | Chem. Sci., 2021, 12, 3161–3169

cyclopentadiene is one of the major initial catalytic pyrolysisproducts responsible for the catalyst deactivation, which shouldbe taken into consideration when optimizing the catalyst andprocess parameters to suppress catalyst deactivation. Ourmechanism may also be generalized to predict the fate of largerlignin moieties and take control of the selectivity. For instance,targeted lignin derivatization to produce catechol while sup-pressing resorcinol and hydroquinone formation may enhancethe conversion and selectivity towards aromatics, such asbenzene.

Conflicts of interest

There are no conicts to declare.

Acknowledgements

The measurements were performed at the VUV beamline of theSwiss Light Source located at Paul Scherrer Institute, Villigen,Switzerland. P. H. and Z. P. are grateful for the funding by SwissNational Science Foundation (SNSF, 200021_178952). A. B., P.H. and Z. P. thank Patrick Ascher for technical assistance. A. P.-U. thanks the Ministry of Science, Technology and Telecom-munications of Costa Rica, and the Costa Rica Institute ofTechnology for their support. All authors want to thank JanaHemberger for the graphical illustrations.

Notes and references

1 C. Li, X. Zhao, A. Wang, G. W. Huber and T. Zhang, Chem.Rev., 2015, 115, 11559–11624.

2 W. Schutyser, T. Renders, S. Van den Bosch, S. F. Koelewijn,G. T. Beckham and B. F. Sels, Chem. Soc. Rev., 2018, 47, 852–908.

3 C. Liu, X. Chen, X. Liu, C. Cui, Z. Zhou, L. Jia and F. Qi,Angew. Chem., Int. Ed., 2021, 60, 2643–2647.

4 P. Azadi, O. R. Inderwildi, R. Farnood and D. A. King,Renewable Sustainable Energy Rev., 2013, 21, 506–523.

5 W. Mu, H. Ben, A. Ragauskas and Y. Deng, BioEnergy Res.,2013, 6, 1183–1204.

6 R. Rinaldi, R. Jastrzebski, M. T. Clough, J. Ralph,M. Kennema, P. C. A. Bruijnincx and B. M. Weckhuysen,Angew. Chem., Int. Ed., 2016, 55, 8164–8215.

7 Z. Ma, E. Troussard and J. A. van Bokhoven, Appl. Catal., A,2012, 423–424, 130–136.

8 G. X. Li, Z. Y. Luo, W. B. Wang and J. M. Cen, Catalysts, 2020,10, 9.

9 A. M. Verma and N. Kishore, RSC Adv., 2017, 7, 2527–2543.10 M. Wang, C. Liu, X. X. Xu and Q. Li, Chem. Phys. Lett., 2016,

654, 41–45.11 J. Zhou, W. Jin, D. K. Shen and S. Gu, J. Anal. Appl. Pyrolysis,

2018, 134, 143–149.12 V. B. F. Custodis, P. Hemberger, Z. Ma and J. A. van

Bokhoven, J. Phys. Chem. B, 2014, 118, 8524–8531.13 A. K. Vasiliou, J. H. Kim, T. K. Ormond, K. M. Piech,

K. N. Urness, A. M. Scheer, D. J. Robichaud,C. Mukarakate, M. R. Nimlos, J. W. Daily, Q. Guan,

© 2021 The Author(s). Published by the Royal Society of Chemistry

Edge Article Chemical Science

Ope

n A

cces

s A

rtic

le. P

ublis

hed

on 1

8 Fe

brua

ry 2

021.

Dow

nloa

ded

on 6

/2/2

021

2:50

:48

PM.

Thi

s ar

ticle

is li

cens

ed u

nder

a C

reat

ive

Com

mon

s A

ttrib

utio

n 3.

0 U

npor

ted

Lic

ence

.View Article Online

H.-H. Carstensen and G. B. Ellison, J. Chem. Phys., 2013, 139,104310.

14 A. M. Scheer, C. Mukarakate, D. J. Robichaud, M. R. Nimlos,H. H. Carstensen and G. B. Ellison, J. Chem. Phys., 2012, 136,044309.

15 D. J. Robichaud, A. M. Scheer, C. Mukarakate, T. K. Ormond,G. T. Buckingham, G. B. Ellison and M. R. Nimlos, J. Chem.Phys., 2014, 140, 234302.

16 V. B. F. Custodis, P. Hemberger and J. A. van Bokhoven,Chem. - Eur. J., 2017, 23, 8658–8668.

17 H. Yang, Y. Furutani, S. Kudo, J.-i. Hayashi and K. Norinaga,J. Anal. Appl. Pyrolysis, 2016, 120, 321–329.

18 Y. Furutani, S. Kudo, J. I. Hayashi and K. Norinaga, J. Phys.Chem. A, 2017, 121, 631–637.

19 M. Gerlach, A. Bodi and P. Hemberger, Phys. Chem. Chem.Phys., 2019, 21, 19480–19487.

20 T. K. Ormond, J. H. Baraban, J. P. Portereld, A. M. Scheer,P. Hemberger, T. P. Troy, M. Ahmed, M. R. Nimlos,D. J. Robichaud, J. W. Daily and G. B. Ellison, J. Phys.Chem. A, 2018, 122, 5911–5924.

21 H. Yang, K. Norinaga, J. Li, W. Zhu and H. Wang, FuelProcess. Technol., 2018, 181, 207–214.

22 P. Hemberger, V. B. F. Custodis, A. Bodi, T. Gerber andJ. A. van Bokhoven, Nat. Commun., 2017, 8, 15946.

23 A. D. Chowdhury and J. Gascon, Angew. Chem., Int. Ed., 2018,57, 14982–14985.

24 Z. W. Fu, Q. R. Shen, C. L. Yao, R. Li and Y. L. Wu,Chemistryselect, 2020, 5, 3011–3017.

25 L. Liguori and T. Barth, J. Anal. Appl. Pyrolysis, 2011, 92, 477–484.

26 A. K. Mondal, C. R. Qin, A. J. Ragauskas, Y. H. Ni andF. Huang, Ind. Crop. Prod., 2020, 148, 8.

27 A. N. Wilson, A. Dutta, B. A. Black, C. Mukarakate,K. Magrini, J. A. Schaidle, W. E. Michener, G. T. Beckhamand M. R. Nimlos, Green Chem., 2019, 21, 4217–4230.

28 A. Bodi, P. Hemberger, D. L. Osborn and B. Sztaray, J. Phys.Chem. Lett., 2013, 4, 2948–2952.

29 P. Hemberger, J. A. van Bokhoven, J. Perez-Ramirez andA. Bodi, Catal. Sci. Technol., 2020, 10, 1975–1990.

30 V. Paunovic, G. Zichittella, P. Hemberger, A. Bodi andJ. Perez-Ramirez, ACS Catal., 2019, 9, 1710–1725.

31 B. Sztaray, K. Voronova, K. G. Torma, K. J. Covert, A. Bodi,P. Hemberger, T. Gerber and D. L. Osborn, J. Chem. Phys.,2017, 147, 013944.

32 M. Johnson, A. Bodi, L. Schulz and T. Gerber, Nucl. Instrum.Methods Phys. Res., Sect. A, 2009, 610, 597–603.

33 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone,

© 2021 The Author(s). Published by the Royal Society of Chemistry

B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato,X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng,J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota,R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery,J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd,E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi,J. Normand, K. Raghavachari, A. Rendell, J. C. Burant,S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam,M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo,J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev,A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski,R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth,P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels,O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski andD. J. Fox, Gaussian 16, Revision A.03, Wallingford CT, 2016.

34 V. A. Mozhayskiy and A. I. Krylov, ezSpectrum, http://iopenshell.usc.edu/downloads.

35 P. Hemberger, Z. Pan, A. Bodi, J. A. van Bokhoven,T. K. Ormond, G. B. Ellison, N. Genossar andJ. H. Baraban, ChemPhysChem, 2020, 21, 2217–2222.

36 A. M. Scheer, C. Mukarakate, D. J. Robichaud, M. R. Nimlos,H.-H. Carstensen and G. B. Ellison, J. Chem. Phys., 2012, 136,044309.

37 M. Frenklach and H. Wang, Proc. Combust. Inst., 1991, 23,1559–1566.

38 J. M. Janssen and D. W. Senser, Combust. Flame, 1991, 84,265–276.

39 H. Wang and M. Frenklach, J. Phys. Chem., 1994, 98, 11465–11489.

40 D. S. Parker, R. I. Kaiser, T. P. Troy and M. Ahmed, Angew.Chem., Int. Ed. Engl., 2014, 53, 7740–7744.

41 V. V. Kislov, N. I. Islamova, A. M. Kolker, S. H. Lin andA. M. Mebel, J. Chem. Theory Comput., 2005, 1, 908–924.

42 R. B. Moffett, Org. Synth., 1952, 32, 41–44.43 G. Zichittella, P. Hemberger, F. Holzmeier, A. Bodi and

J. Perez-Ramirez, J. Phys. Chem. Lett., 2020, 11, 856–863.44 J. R. Anderson, F. Chang and R. J. Western, J. Catal., 1989,

118, 466–482.45 A. Sato, T. Miyake and K. Sekizawa, Appl. Catal., A, 1994, 111,

1–9.46 R. Xu, J. N. Jocz, L. K. Wiest, S. C. Sarngadharan, M. Milina,

J. S. Coleman, L. L. Iaccino, P. Pollet, C. Sievers andC. L. Liotta, Ind. Eng. Chem. Res., 2019, 58, 22516–22525.

47 M. Diedhiou, B. J. West, J. Bouwman and P. M. Mayer, J.Phys. Chem. A, 2019, 123, 10885–10892.

48 F. J. Weigert, J. Org. Chem., 1987, 52, 921–923.

Chem. Sci., 2021, 12, 3161–3169 | 3169